Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Questions about data assimilation, nudging, and whether I need WRFDA

This post was from a previous version of the WRF&MPAS-A Support Forum. New replies have been disabled and if you have follow up questions related to this post, then please start a new thread from the forum home page.


New member

I took the WRF winter tutorial this past January, and by now have had some experience
with running WPS/WRF simulations for a geographical region where we will be doing some field tests.
Ultimately we want to get information about the state of the atmosphere in order
to model RF propagation over the ocean.

We want to be able to incorporate point measurements of meteorological data,
which will be available sporadically at one or a few times during a simulated period,
to improve the characterization of the atmosphere.

I think I understand that this could be done with observation nudging,
which was described briefly in one section of the tutorial,
but the more I read the more confused I am.

I originally had the idea that there was a distinction between “data assimilation”
which was to incorporate sets of data available periodically at several times during a simulation,
into the meteorological inputs and initial and boundary conditions;
vs “nudging” (specifically observation nudging) which added some forcing terms
to adjust the model outputs based on locally available weather data not originally input
to the preprocessor.

However, based on what I’ve been reading, it seems that “nudging” is the method behind the data assimilation?
It is mentioned in connection with FDDA (Four-Dimensional Data Assimilation I think?).
I'm not sure whether FDDA requires/implies WRFDA or not.
So I have several questions.

First, is it possible to do what I had in mind, i.e. use sporadically available meteorological data inputs
to adjust the model output?

Next, to what extent does that have to be set up in WPS, or is it done through setups to WRF
(real.exe, wrf.exe and associated routines)?

Next, do I need OBSGRID for this? I get the idea that it’s primarily
for large fields of satellite data covering the simulation area,
and is done along with/right after the WPS steps.

Next and perhaps most importantly, do I need to build WRFDA in order to do observation nudging?
FDDA was covered in one class of the WRF tutorial, and this included grid nudging
(which I think isn’t what we need), as well as surface nudging and observation nudging,
but there was no mention of needing to build/use WRFDA in the class that I recall.
But if I do need to build and use WRFDA, then I also need to sign up for the tutorial,
which is less than two weeks from now.
So I need to determine whether it’s necessary right now.

Thanks in advance for advice on this.

Fred Hawes

Based on my understanding of your question, I don't think you need to run WRFDA, which is for variational data assimilation. For your case, you may need to run observational nudging, which is one component of WRF-ARW. Please see details in chapter 7 of the WRF user guide (

However, note that observation nudging works fine when you have dense observations. If you only have sporadic in-situ observations, they might not be sufficient enough to improve the simulation.

Thank you for the information, and for the quick reply. That was quite helpful.

We also have another application in the future, in which we will have weather data from a Raman Lidar,
which will generate 3-D profiles for atmospheric temperature and water vapor content continually,
at some scanning periodicity. This is a much denser set of observation data.
Would that still be the sort of case that WRF can handle without WRFDA?

Fred Hawes
You are right. If you have high-quality, dense observations, you can always run observational nudging to incorporate them into WRF.